1 23

Demography ISSN 0070-3370Volume 49Number 2 Demography (2012) 49:575-606DOI 10.1007/s13524-012-0096-x

Widowhood and Mortality: A Meta-Analysis and Meta-Regression

Eran Shor, David J. Roelfs, MistyCurreli, Lynn Clemow, Matthew M. Burg& Joseph E. Schwartz

1 23

Your article is protected by copyright and

all rights are held exclusively by Population

Association of America. This e-offprint is

for personal use only and shall not be self-

archived in electronic repositories. If you

wish to self-archive your work, please use the

accepted author’s version for posting to your

own website or your institution’s repository.

You may further deposit the accepted author’s

version on a funder’s repository at a funder’s

request, provided it is not made publicly

available until 12 months after publication.

Widowhood and Mortality: A Meta-Analysisand Meta-Regression

Eran Shor & David J. Roelfs & Misty Curreli &Lynn Clemow & Matthew M. Burg &

Joseph E. Schwartz

Published online: 17 March 2012# Population Association of America 2012

Abstract The study of spousal bereavement and mortality has long been a major topicof interest for social scientists, but much remains unknown with respect to importantmoderating factors, such as age, follow-up duration, and geographic region. The presentstudy examines these factors using meta-analysis. Keyword searches were conducted inmultiple electronic databases, supplemented by extensive iterative hand searches. Weextracted 1,377 mortality risk estimates from 123 publications, providing data on morethan 500 million persons. Compared with married people, widowers had a mean hazardratio (HR) of 1.23 (95% confidence interval (CI), 1.19–1.28) among HRs adjusted for

Demography (2012) 49:575–606DOI 10.1007/s13524-012-0096-x

Electronic supplementary material The online version of this article (doi:10.1007/s13524-012-0096-x)contains supplementary material, which is available to authorized users.

E. ShorDepartment of Sociology, McGill University, Room 713, Leacock Building, 855 Sherbrooke StreetWest, Montreal, Quebec H3A 2T7, Canadae-mail: ershor@gmail.com

D. J. Roelfs (*)Department of Sociology, University of Louisville, Louisville, KY 40292, USAe-mail: david.roelfs@louisville.edu

M. CurreliDepartment of Sociology, Stony Brook University, Stony Brook, NY 11794-4356, USAe-mail: misty.curreli@stonybrook.edu

L. Clemow :M. M. BurgBehavioral Cardiovascular Health and Hypertension Program, Columbia University Medical Center,622 West 168th Street, New York, NY 10032, USAe-mail: lc2193@columbia.edu

M. M. Burge-mail: mb2358@columbia.edu

J. E. SchwartzDepartment of Psychiatry and Behavioral Science, Stony Brook University,Stony Brook, NY 11794-8790, USAe-mail: jschwartz@notes.cc.sunysb.edu

Author's personal copy

age and additional covariates and a high subjective quality score. The mean HR washigher for men (HR, 1.27; 95% CI, 1.19–1.35) than for women (HR, 1.15; 95% CI,1.08–1.22). A significant interaction effect was found between gender and mean age,with HRs decreasing more rapidly for men than for women as age increased. Othersignificant predictors of HRmagnitude included sample size, geographic region, level ofstatistical adjustment, and study quality.

Keywords Widowhood .Marital status . Mortality . Meta-analysis

Introduction

The effect of marital status on health and mortality was one of the earliest issues to besystematically studied by sociologists and demographers, with work dating to Durkheim’sclassic study on suicide (Durkheim 1897/1951). Over the years, numerous studies haveexamined this relationship, with many of them focusing on the risk of death amongpersons who had lost their spouse (e.g., Alter et al. 2007; Clayton 1974; Hart et al.2007; Helsing et al. 1981; Jones and Goldblatt 1987; Lusyne et al. 2001; Manor andEisenbach 2003; Schaefer et al. 1995; Stimpson et al. 2007; Young et al. 1963).Indeed, the death of a loved one is widely recognized as one of life’s most potentstressors, due in part to the associated disruption of social support, life routines, andfinancial status (Stroebe 2001).

Overall, the resulting body of research demonstrates a higher risk of deathassociated with loss of a spouse (Hughes and Waite 2009), although a few studiesreport no significant effect and the magnitude of effect varies substantially. Thisvariability is at least partly associated with individual factors that include gender(Mineau et al. 2002; Schaefer et al. 1995; Smith and Zick 1996; Stroebe et al. 2001;Thierry 1999), age (Johnson et al. 2000; Lichtenstein et al. 1998; Manor andEisenbach 2003; Martikainen and Valkonen 1996a; Mendes De Leon et al. 1993;Schaefer et al. 1995), recency of widowhood (Jagger and Sutton 1991; Kaprio et al.1987; Mellstrom et al. 1982; Nystedt 2002; Stimpson et al. 2007; Stroebe et al. 2007),and geographical region (Lusyne et al. 2001; Nagata et al. 2003; Rahman et al. 1992;Voges 1996).

The current trend in the literature is toward an increased emphasis on identifyingmediating, moderating, and confounding factors in the widowhood/mortality association.Thus, the time is ripe for a meta-analysis that examines known potential moderators andseeks to identify new ones.

Moderating Factors in the Widowhood-Mortality Association

In their recent meta-analysis of the literature on marital status (including widowhood)and mortality among individuals 65 years of age and older, Manzoli et al. (2007)reported that widowed persons had an 11% higher risk of mortality compared withmarried persons. Moderating factors such as gender and geographic region, however,were nonsignificant. The former finding is surprising given that gender differences inmarital status–related mortality have been well established by previous studies (Gove1973; Hemstrom 1996; Stroebe et al. 2001). As typical examples, it has been found

576 E. Shor et al.

Author's personal copy

that relative risks (widowhood vs. married) among men were 16% higher (Hemstrom1996) to 42% higher (Kolip 2005) than relative risks among women. It is interestingto note, however, that gender differences such as these tend to be found in non-elderlysamples rather than among the older cohorts examined by Manzoli et al. (2007). Thissuggests the possibility of a gender-age interaction, an idea that is supported bystudies examining both gender and age. Mineau et al. (2002) found that the mortalityrisk of widowed men relative to married men, compared with the same measureamong women, was 46% higher at age 35–44 but only 12% higher at ages 75 andolder. Even more dramatically, Smith and Zick (1996) found that the men’s relativerisk was 373% higher at ages 25–64 but 22% lower among those 65 and older.

Other potentially important moderators have also received attention. First, studieshave considered the possible effect of time period on the magnitude of thewidowhood-mortality association. Despite steady improvements in medical treat-ments, some studies found that widowhood-related relative risks have increased overtime. For example, van Poppel and Joung (2001) found increases in relative risks of43% among men and 87% among women in the Netherlands between the 1850–1859and the 1960–1969 periods. Unexplained increases, ranging from 6% to 40%, havealso been found among Finnish men and women between the 1976–1980 and 1996–2000 periods (Martikainen et al. 2005). However, Mineau et al. (2002) found bothincreases and decreases in relative risks between an 1860–1874 and an 1895–1904marriage cohort, depending on age at widowhood. In this study, relative risksincreased by between 12% and 19% among persons widowed between the ages of25 and 44, remained unchanged for ages 45–64, and declined by between 6% and10% for persons widowed at age 65 or older. In their comparison of Sweden,Belgium, and the Netherlands, Alter et al. (2007) also showed decreased relativerisks over time for women who were widowed less than 5 years.

Second, the possible effects of widowhood recency have been central to a long lineof research. At an early date, Young and colleagues noted that mortality was highestin the first six months following widowhood but declined in subsequent months andyears (Young et al. 1963). Subsequent research has largely supported this finding. Forexample, Nystedt (2002) found that widowhood-related relative risks (RRs) fell from2.38 in the first six months of widowhood to 1.28 among those for whom widowhoodoccurred six or more years prior. Likewise, Thierry (1999) found that RRs (widowedvs. married) fell during the first 10 years for men and women of all ages.

The suggestion that the risk of mortality varies depending on the amount of timeelapsed from the onset of widowhood has opened the way for physiological inves-tigations of loss and grief from a stress response perspective (Jones and Goldblatt2006; Martikainen and Valkonen 1996b; Susser 1981). In doing so, these studies havefocused on linkage mechanisms, such as immune system disruption (Gerra et al.2003; Goforth et al. 2009) and cardiovascular effects (Buckley et al. 2010), that mayconnect the early months of widowhood to a range of chronic diseases and mortality.Others have examined behavioral pathways, such as poor self-care and increases inhealth-risk behaviors by the surviving spouse, that potentially connect bereavementto near-term negative health consequences (Jin and Christakis 2009; Sharar et al.2001; Stroebe et al. 2007). A related line of research has focused on the loss ofimportant social (Armenian et al. 1987; Bowling and Charlton 1987; Jylha and Aro1989; Lusyne et al. 2001; Martikainen and Valkonen 1996b; Mineau et al. 2002) and

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 577

Author's personal copy

economic (Nystedt 2002; Rahman 1997; Smith and Waitzman 1994; Zick and Smith1991b) buffers that may affect health and survival (Subramanian et al. 2008).

The present meta-analysis contributes to this body of knowledge on widowhoodand mortality in two important ways. First, we utilize the heterogeneity of researchsettings found in this literature to assess the impact of multiple potential moderators.Some, such as gender and age, are relatively easy to evaluate within an individualstudy. Others—such as widowhood recency, time period, and cultural differences—are less frequently addressed by individual studies and are therefore more easilyexamined by comparing across studies. Meta-analysis is well suited to this task, andour results include tests of gender-age interactions, geographic region, time period,and a number of specific study-design characteristics. Second, the overall magnitudeof the association between widowhood and health outcomes has not been examinedamong non-elderly persons.

Specifically, we test four hypotheses using meta-analysis and meta-regression.First, we assess whether there exists a gender-age interaction such that hazard ratios(HRs) are greater for men than for women, but more so at younger ages. Second, wetest the hypothesis that the relative mortality risk associated with widowhood hasincreased over time. Third, we examine the hypothesis put forth by Young et al.(1963) that more recently experienced widowhood is associated with greater mortal-ity risk. Finally, we test whether there exists an interaction between gender andfollow-up duration such that HRs are greater for men than for women, particularlywhen widowhood is recent. In addition, we examine the possible effects of geograph-ic region, study-level control variables, the composition of the case and controlgroups, and several other study-design characteristics in the interest of providingfindings from which new mediator and moderator hypotheses might be developed.

Methods

In June 2005, we conducted a sensitive search of electronic bibliographic databases toretrieve all publications combining the concepts of psychosocial stress (includingwidowhood) and all-cause mortality. Overall, 100 search clauses were used for Medline,97 for EMBASE, 81 for CINAHL, and 20 for Web of Science. (See Online Resource 1,Section 1, for the full search algorithm used for Medline; information on the remain-ing search algorithms is available from the authors upon request.) This processidentified 1,570 unique publications. With these results as a base, the bibliographiesof eligible publications, the lists of sources citing an eligible publication, and thesources identified as “similar to” an eligible publication were iteratively hand-searched. The literature was exhausted after eight iterations. (The full description ofthis iterative search protocol is available from the authors upon request.). Theelectronic keyword searches in these databases were run again in July 2008, andthe search and coding stages were completed in January 2009.

The electronic database searches were performed by a research librarian. Twoauthors trained in meta-analysis coding procedures determined publication eligibilityand extracted the data from the identified articles. Data were entered into andpublications were tracked throughout the process using spreadsheets. (See OnlineResource 1, Section 2, for a full list of variables for which data were sought.) All

578 E. Shor et al.

Author's personal copy

unpublished work encountered was considered for study inclusion. Although thesearch was done for articles published in English, we were able to locate and translatethe relevant portions of 36 publications in German, Danish, French, Spanish, Dutch,Polish, and Japanese. Figure 1 summarizes the number of publications considered ateach step of the search process. Among the 730 publications considered tentativelyeligible for study inclusion (based on examinations of title only), 428 were excludedfrom further consideration upon examination of the abstract. Of the remaining 302publications that were examined in full, 151 were excluded because of the lack of avalid stress measure (70 publications), unavailable data on the case or control group(37 publications), lack of the all-cause mortality outcome (15 publications), confla-tion of multiple stressors (13 publications), the lack of a valid comparison group (11publications), and for other reasons (6 publications). The full database contains 262publications examining the effects of various stressful events on all-cause mortality.To evaluate coding accuracy, 40 of these publications were randomly selected andrecoded (including 446 point estimates). Of the point estimates, 98.6% were errorfree.

The present analysis uses the subset of articles (n 0 123) that reported theeffect of widowhood on all-cause mortality. Of these publications, 116 appeared inpeer-reviewed journals, four were book chapters, one was an unpublished disserta-tion, and two were unpublished papers (authors of these papers were contacted for

123 publications included in meta-analysis of widowhood and mortality

139 publications containing stress measures other than widowhood

12,448 titles citing an eligible publication

About 12,000 titles identified in

bibliographies of coded publications

1,570 publications identified by

original keyword search

About 50,000 titles identified as

“similar to” an eligible publication

455 publications tentatively met study inclusion

criteria

227 publications tentatively met study inclusion

criteria

48 publications tentatively met study inclusion

criteria

305 excluded

145 excluded

18 excluded

150 coded

82 coded

30 coded

Total pool of 262 publications for meta-analyses of stress and all-cause mortality

Keyword Search Hand Search

Fig. 1 Flow chart of publications reviewed for study eligibility

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 579

Author's personal copy

permission to use their results). One publication was translated from Spanish, twofrom German, one from French, and one from Danish in consultation with fluentspeakers of the language; the remaining 118 publications were in English (seeTable 1).

In addition to the requirement that a study report one or more point estimatespertaining to all-cause mortality, studies were included in the present analysis if aclear comparison was made between people who lost their spouse and people whowere married (or the general population, which consists primarily of married people;see Table 1). In addition to studies with longitudinal designs, cross-sectional studieswere included if the sample size was large, a baseline date could be determinedaccurately, and the manner in which death data was collected approximated a follow-up period. For example, the study by Sheps (1961), which used census data for thedenominator and national annual mortality data for the numerator, was coded ashaving an April 1, 1950, baseline and a one-year follow-up period. In total, the 123publications provided 1,377 point estimates for analysis.

Statistical methods varied across the included studies, necessitating the conversionof odds ratios, rate ratios, standardized mortality ratios, RRs, and HRs into a commonmetric. All non-HR point estimates were converted to HRs (the most frequentlyreported type) using one or both of the following equations (Zhang and Yu 1998):

RR ¼ OR

1� rð Þ þ r � ORð Þ

and

HR ¼ ln 1� RR � rð Þln 1� rð Þ ;

where RR is the relative risk, OR is the odds ratio, HR is the hazard ratio, and r is thedeath rate for the reference group (i.e., those who are married). For 328 of the 1,377relative risks, the death rate (i.e., the conversion factor, not the dependent variable)for the reference group was not reported. In these cases, the death rate was estimatedusing multiple regression. Significant predictors of the death rate were follow-upduration, sample size (log transformed), the proportion of the sample that was male,mean age at enrollment, an indicator for whether the study statistically controlled forgender, an indicator for whether the study statistically controlled for age, and thesubjective quality assessment score assigned by the coders (multiple R 0 .797).Sensitivity analyses were performed to examine the possible effect of including orexcluding studies for which we had to estimate the death rate.

As is standard practice, the standard errors reported in the publications were usedto calculate the inverse variance weights. When not reported, standard errors werecalculated using (1) confidence intervals, (2) t statistics, (3) chi-square statistics, or(4) p values. When upper-limit p values were the only estimate of statistical signif-icance available (e.g., when the reported p value was between .01 and .05), themidpoint of the upper and lower limits was used to estimate the true p value. For 668of the 1,377 point estimates, no measure of statistical significance was reported andstandard errors were estimated using multiple regression. Significant predictors of thestandard error were follow-up duration, sample size (log transformed), mean age at

580 E. Shor et al.

Author's personal copy

Tab

le1

Studies

included

inthemeta-analyses

andmeta-regressions

Pub

lication

DataSource

Cou

ntry

Reference

Group

Years

Sam

pleSize

MeanHR

No.

ofHRs

Alteret

al.(200

7)HSN

Netherlands

Married

1850–188

91,973

1.06

9

OriginalData

Belgium

1811–190

0

SDD

Sweden

1766–189

5

Arm

enianet

al.(198

7)AAOC

Lebanon

Married

1949–198

03,058

1.46

12

Bagiella

etal.(200

5)EPESE

US

Married

1981–200

214

,456

1.16

6

Ben-Shlom

oet

al.(199

3)WhitehallStudy

UK

Married

1967–199

018

,403

1.50

3

Berkson

(196

2)Census,19

50US

Married

1949–195

115

0,52

0,79

81.50

2

Bow

lingandBenjamin

(198

5)OPCSLS

UK

General

popu

latio

n19

79–198

450

30.85

24

Breezeet

al.(199

9)OPCSLS

UK

Married

1971–199

293

,931

1.06

8

BrockmannandKlein

(200

2)GSOEP

Germany

Married

1984–199

818

,538

1.09

16

BrockmannandKlein

(200

4)GSOEP

Germany

Married

1984–199

812

,484

1.29

8

Burgo

aet

al.(199

8)Census,19

91Spain

Married

1991–199

138

,939

,050

1.80

4

Cam

pbellandLee

(199

6)EBHR

China

Married

1792–199

312

,000

1.52

6

Cheung(200

0)HLSS

UK

Married

1984–199

73,378

1.11

2

ChristakisandAllison(200

6)Medicare

US

Married

1993–200

251

8,24

01.37

4

Clayton

(197

4)OriginalData

US

Married

1968–196

921

80.80

2

Com

stockandTo

nascia

(197

7)Health

Census

US

Married

1963–197

147

,423

2.01

2

Dob

lham

mer

(200

0)Census,1981

Austria

Married

1981–199

71,254,15

31.24

1

Dzurova

(200

0)Census,19

90,19

95Czech

Rep

Married

1990–199

510

,306

,000

2.39

2

Ebrahim

etal.(199

5)BRHS

UK

Married

1978–199

07,735

1.32

6

Ekb

lom

(196

3)Census,19

51–195

8Sweden

General

popu

latio

n19

51–196

163

41.24

15

ElwertandChristakis(200

8)Medicare

US

Married

1993–200

274

6,37

81.17

2

EspinosaandEvans

(200

8)NLMS

US

Married

1979–200

572

,242

1.46

4

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 581

Author's personal copy

Tab

le1

(contin

ued)

Pub

lication

DataSource

Cou

ntry

Reference

Group

Years

Sam

pleSize

MeanHR

No.

ofHRs

Goldm

anandHu(199

3)Census,19

75Japan

Married

1975–198

5111,940,000

1.44

2

Goldm

anet

al.(199

5)NHIS-SA

US

Married

1984–199

07,478

1.14

2

Goo

dwin

etal.(198

7)NMTR

US

Married

1969–198

225

,706

1.17

1

GrundyandKravdal

(200

8)Census,19

80Norway

Married

1980–200

31,530,10

11.51

6

Hajdu

etal.(199

5)Census,19

69–198

9Hun

gary

Married

1969–199

115

,486

,537

1.86

40

UK

Hartet

al.(200

7)R-PS

UK

Married

1972–200

48,790

1.14

15

HaywardandGorman

(200

4)NLS-O

MUS

Married

1966–199

04,562

1.37

1

Helsing

andSzklo

(198

1)Health

Census

US

Married

1963–197

58,064

1.30

57

Helsing

etal.(198

1)Health

Census

US

Married

1963–197

58,064

1.30

14

Helweg-Larsenet

al.(200

3)DANCOS

Denmark

Married

1987–199

96,693

1.11

1

Hem

strom

(199

6)Census,19

80Sweden

Married

1980–198

61,896,62

61.37

2

Henretta

(200

7)HRS

US

Married

1994–200

24,335

1.39

1

Horwitz

andWeber

(197

4)Census,1968

Denmark

Married

1968–196

81,710,80

02.00

20

Ikedaet

al.(200

7)JC

CS

Japan

Married

1988–199

990

,064

1.39

10

Iribarrenet

al.(200

5)CARDIA

US

Married

1985–200

05,115

1.76

1

Iwasakiet

al.(200

2)Kom

o-IseStudy

Japan

Married

1993–200

011,565

1.14

4

Jagg

erandSutton(199

1)OriginalData

UK

Married

1980–198

834

41.35

2

Jenk

insonet

al.(199

3)ASSET

UK

Married

1986–199

01,376

1.64

3

Johansen

etal.(199

6)DCR

Denmark

Married

1968–199

47,302

1.35

4

Johnsonet

al.(200

0)NLMS

US

Married

1978–198

928

1,46

01.28

20

JonesandGoldb

latt(198

7)OPCS-LS

UK

General

popu

latio

n19

71–198

126

4,28

41.18

20

Joneset

al.(198

4)

OPCS-LS

UK

General

popu

latio

n19

71–197

626

4,28

41.23

17

582 E. Shor et al.

Author's personal copy

Tab

le1

(contin

ued)

Pub

lication

DataSource

Cou

ntry

Reference

Group

Years

Sam

pleSize

MeanHR

No.

ofHRs

Jounget

al.(199

6)Census,19

86Netherlands

Married

1986–199

014

,572

,000

1.26

2

JylhaandAro

(198

9)OriginalData

Finland

Married

1979–198

51,060

1.17

4

Kalediene

etal.(200

7)Census,19

89,20

01Lith

uania

Married

1989–200

13,670,00

01.65

4

KaplanandKronick

(200

6)NHIS

US

Married

1989–199

780

,018

1.35

1

Kaprioet

al.(198

7)Census,19

72Finland

General

popu

latio

n19

72–197

695

,647

1.07

1

Keller(196

9)OriginalData

US

Married

1959–196

670

60.62

2

Koh

lerandKoh

ler(200

2)DanishTw

inRegistry

Denmark

Married

1968–200

07,093

1.06

4

Kolip

(200

5)Census,1999

Germany

Married

1999–200

045

,500

,000

2.08

2

Kraus

andLilienfeld

(195

9)Census,19

50US

Married

1949–195

115

0,52

0,79

81.94

32

Kravdal

(200

3)Census,19

60–199

0Norway

Married

1960–199

931

,998

1.18

3

Kravdal

(200

7)Census,19

80–199

0Norway

Married

1980–199

94,091,00

01.23

8

Kroenke

etal.(200

6)Nurses'Health

Study

US

Married

1992–200

42,835

0.95

2

Kum

leandLun

d(200

0)Census,19

70Norway

Married

1970–198

91,338,71

61.20

7

Lichtensteinet

al.(199

8)STR

Sweden

Married

1981–199

339

,846

1.32

31

LillardandPanis(199

6)PSID

US

Married

1984–199

04,092

1.47

2

Litw

in(2007)

Census,19

97Israel

Married

1997–200

41,811

1.26

2

Liu

andSulliv

an(200

3)OriginalData

US

Married

1994–199

864

61.36

2

Lusyn

eet

al.(200

1)Census,19

91Belgium

Married

1991–199

635

3,19

01.27

10

Makuc

etal.(199

0)NHANESI

US

Married

1971–198

43,826

1.12

2

Malyu

tinaet

al.(200

4)MONICA,Nov

osibirsk

Russia

Married

1984–199

811,404

2.41

6

Manor

andEisenbach

(200

3)ILMS

Israel

Married

1983–199

290

,830

1.42

59

Manor

etal.(199

9)ILMS

Israel

Married

1983–199

272

,527

1.10

5

Manor

etal.(200

0)ILMS

Israel

Married

1983–199

279

,623

1.08

4

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 583

Author's personal copy

Tab

le1

(contin

ued)

Pub

lication

DataSource

Cou

ntry

Reference

Group

Years

Sam

pleSize

MeanHR

No.

ofHRs

Martelin

(199

6)Census,19

70Finland

Married

1970–197

54,606,30

71.15

4

Martelin

etal.(199

8)Census,19

70–198

5Finland

Married

1970–199

04,606,30

71.09

4

Martik

ainen(199

0)Census,19

80Finland

Married

1980–198

54,779,53

52.06

1

Martik

ainen(199

5)Census,19

80Finland

Married

1980–198

54,779,53

51.18

2

Martik

ainenandValkonen(199

6a)

Census,19

85Finland

Married

1985–199

11,580,00

01.18

6

Martik

ainenandValkonen(199

6b)

Census,19

85Finland

Married

1985–199

11,580,00

01.20

12

Martik

ainenandValkonen(199

8)Census,19

85Finland

Married

1985–199

14,901,78

31.37

4

Martik

ainenet

al.(200

5)Census,19

75,19

95Finland

Married

1975–200

03,165,00

01.25

24

Mellstrom

etal.(198

2)Census,19

68–197

8Sweden

Married

1968–197

87,914,00

00.93

48

MendesDeLeonet

al.(199

2)KRIS

Netherlands

Married

1972–198

23,365

1.25

3

MendesDeLeonet

al.(199

3)YHAP

US

Married

1982–198

81,046

1.75

28

Metayer

etal.(199

6)DCS

US

Married

1990–199

313

82.03

3

Mineauet

al.(200

2)UPDB

US

Married

1860–198

962

,336

1.10

44

Mollicaet

al.(200

1)OriginalData

Croatia

Married

1996–199

952

91.99

1

Mostafa

andvanGinneken(200

0)MatlabDSS

Bangladesh

Married

1982–199

210

,000

1.64

4

Nagataet

al.(200

3)TakayamaStudy

Japan

Married

1992–199

93,505

0.71

10

Neale

etal.(198

6)OriginalData

US

Married

1949–196

81,261

1.56

1

Niemi(197

9)CPSI

Finland

Married

1964–197

693

90.86

3

Nilssonet

al.(200

5)MPP

Sweden

Married

1974–199

253

,111

1.11

4

Nyb

oet

al.(200

3)Census,1998

Denmark

Married

1998–200

02,249

1.03

4

Nystedt

(200

2)SDD

Sweden

Married

1800–189

51,252

1.58

12

Okamotoet

al.(200

7)OriginalData

Japan

Married

1995–200

178

41.61

4

Ortmeyer

(197

4)Census,19

60US

Married

1959–196

118

0,67

1,00

01.72

4

584 E. Shor et al.

Author's personal copy

Tab

le1

(contin

ued)

Pub

lication

DataSource

Cou

ntry

Reference

Group

Years

Sam

pleSize

MeanHR

No.

ofHRs

Parkeset

al.(196

9)NHS

UK

Married

1957–196

64,486

1.40

1

Peritz

etal.(196

7)Census,19

49–196

2Israel

Married

1949–196

22,150,00

01.35

48

Rahman

(199

3)MatlabDSS

Bangladesh

Married

1974–198

264

,263

1.25

10

Rahman

(199

7)MatlabDSS

Bangladesh

Married

1974–198

224

,889

1.82

16

Rahman

etal.(199

2)MatlabDSS

Bangladesh

Married

1974–198

224

,889

1.12

6

RasuloandChristensen

(200

4)LSADT

Denmark

Married

1995–200

152

41.07

6

ReesandLutkins

(196

7)OriginalData

UK

Married

1960–196

61,781

2.03

10

Regidor

etal.(200

1)Census,19

96Spain

Married

1996–199

83,110,121

2.20

12

Rosengren

etal.(198

9)GMPPT

Sweden

Married

1970–198

39,869

1.80

2

Ryan(199

2)OriginalData

UK

Married

andGeneral

populatio

n19

71–198

945

50.77

21

Sam

uelssonandDehlin

(199

3)OriginalData

Sweden

Married

1922–199

039

20.53

2

Schaeferet

al.(199

5)KFHP

US

Married

1964–198

712

,522

1.46

34

Sheps

(196

1)Census,19

50US

Married

1949–195

1112,354,03

41.99

36

Shk

olnikovet

al.(200

7)Census,20

01Lith

uania

Married

2001–200

43,481,29

51.69

4

Shu

rtleff(195

5)Census,19

40,19

50US

Married

1940–195

115

0,52

0,79

81.74

22

Shu

rtleff(195

6)Census,19

50US

Married

1950–195

115

0,52

0,79

81.31

8

Silv

ersteinandBengtson(199

1)U.S.C.LSG

US

Married

1971–198

543

92.62

1

Singh

andSiahp

ush(200

1)NLMS

US

Married

1979–198

930

1,18

32.45

6

Singh

andSiahp

ush(200

2)NLMS

US

Married

1979–198

930

0,91

01.09

3

Smith

andWaitzman

(199

4)NHANESI

US

Married

1971–198

420

,729

1.23

8

Smith

andZick(199

6)PSID

US

Married

1968–198

23,564

0.79

24

Sorlie

etal.(199

5)NLMS

US

Married

1979–198

953

0,50

71.50

12

Spence(200

6)NLS-M

WUS

Married

1967–200

13,258

1.18

1

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 585

Author's personal copy

Tab

le1

(contin

ued)

Pub

lication

DataSource

Cou

ntry

Reference

Group

Years

Sam

pleSize

MeanHR

No.

ofHRs

Stim

pson

etal.(200

7)EPESE

US

Married

1993–200

01,693

1.75

4

Sub

ramanianet

al.(200

8)Medicare

US

Married

1993–200

240

0,00

01.17

2

Thierry

(199

9)Census,19

69–197

4France

Married

1969–199

656

,615

,155

2.21

158

Census,19

89–199

1

Tomassini

etal.(200

1)LSADT

Denmark

Married

1997–199

92,172

1.40

1

Vallin

andNizard(197

7)Census,19

68France

Married

1967–196

949

,780

,543

1.99

30

vanPoppelandJoun

g(200

1)Census,18

50–196

0Netherlands

Married

1850–196

911,486,630

1.25

24

Villingshojet

al.(200

6)

DCR

Denmark

Cohabitatin

g1991–200

277

00.96

7

Vog

es(199

6)GSOEP

Germany

Married

1984–199

33,699

1.04

3

You

nget

al.(196

3)Ministryof

Health

UK

General

popu

latio

n19

57–196

04,486

1.12

43

Zajacov

a(200

6)NHANESI

US

Married

1971–199

212

,036

1.19

2

ZickandSmith

(199

1a)

PSID

US

Married

1971–198

41,990

1.22

4

586 E. Shor et al.

Author's personal copy

enrollment, the magnitude of the HR, and publication date (multiple R 0 .721). Anindicator variable was created so that analyses could be conducted both with andwithout the estimated standard errors.

Many meta-analysts prefer to use only the most general point estimates reported ina given publication. Although this strategy makes it easier to maintain independencebetween point estimates and makes the calculations of the inverse variance weightsstraightforward, it also results in a substantial loss of information. We sought, instead,to maximize the number of point estimates analyzed, capturing variability bothbetween publications and within each publication rather than just the former. Forexample, when a publication (see hypothetical Study X in Table 2) reported mortalityrisks by gender subgroups alone, the data require no adjustment. Likewise, when astudy reported mortality risks by age group alone (see hypothetical Study Y, Table 2),the data again require no adjustment. When a publication first reported mortality risksby gender and then again by age, however (see hypothetical Study Z, Table 2), thiscreated a violation of independence because each person is represented twice. Tocorrect for this double-counting, each of the variance weights was adjusted to one-half of its original value, thus preserving information on the gender and age variableswhile counting each subject only once.

Two measures of study quality were adopted. First, a three-level subjective rating wasassigned to each publication. Publications were rated as “low quality” if they containedobvious reporting errors or applied statistical methods incorrectly. Publications were ratedas “high quality” if models were well specified (i.e., the correct model was used relative tothe state of the art at the time of publication) and if discussions and reporting of study resultswere detailed. Next, principal components analysis was used to construct a 10-point scaleusing the following: (1) the five-year impact factor (ISI Web of Knowledge 2009) of thejournal in which the article was published (an impact factor of 1 was assigned whenthe impact factor was not available); (2) the number of citations received per yearsince publication according to ISI Web of Knowledge; and (3) the number of authors,since studies with a larger author body may have a more diverse pool of scholarlyexpertise, thereby decreasing the likelihood of methodological or theoretical errors.

Table 2 Illustration of adjustments made to the inverse variance weights to correct for double reporting

Author, Publication Year Gender AgeOriginal InverseVariance Weight

Corrected InverseVariance Weight

Study X Men only All ages 4 4

Study X Women only All ages 2 2

Study Y Men only 20–44 5 5

Study Y Men only 45–65 7 7

Study Y Men only 65+ 3 3

Study Z Men only All ages 12 6

Study Z Women only All ages 20 10

Study Z Both men and women 20–44 16 8

Study Z Both men and women 45–65 24 12

Study Z Both men and women 65+ 16 8

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 587

Author's personal copy

Both Q tests (which assess the probability that the observed variability amongeffect estimates, across studies and/or subgroup, within a meta-analysis is due solelyto chance) and I2 tests (which use the results of the Q test to calculate the degree ofheterogeneity present) were used to assess heterogeneity in the data (Huedo-Medinaet al. 2006). Q test results from preliminary analyses revealed substantialheterogeneity. In light of this, all meta-analyses and meta-regression analyses werecalculated by maximum likelihood using a random-effects model, the random effectsbeing applied at the level of the HR data. Analysis was performed with PASWStatistics 18.0 using matrix macros provided by Lipsey and Wilson (2001). Thepossibility of selection and publication bias was examined using a funnel plot ofthe log HRs against sample size. Because of heterogeneity in the data, funnel plotasymmetry was tested using Eggers’ test (Egger and Davey-Smith 1998) and weight-ed least squares regressions of the log HRs on the inverse of the sample size (Morenoet al. 2009; Peters et al. 2006).

Analyses performed include meta-analyses of subgroups and multivariate meta-regression analyses. The covariates used in the analyses were dictated by data availabil-ity. Variables such as race or ethnicity, which were used as grouping variables orincluded in interaction terms in only a small number of studies, could not be used inthe analyses. The following covariates were used: (1) whether the death rate had to beestimated in order to derive the HR (yes or no); (2) whether the standard error wasestimated (yes or no); (3) the proportion of respondents whowere male; (4) the mean ageof the sample/subgroup at enrollment, divided by 10; (5) the age range of the sample/subgroup at enrollment, divided by 10; (6) age of the study (the years elapsed since thebeginning of the enrollment period), divided by 10; (7) age of the publication (the yearselapsed since publication), divided by 10; (8) the duration of the enrollment period, inyears; (9) the time elapsed between the end of enrollment and the beginning of follow-up, in years; (10) the follow-up duration, in years; (11) whether the general populationwas used as the comparison group (yes or no), as opposed to only married persons; (12)whether the study sample consisted of persons with preexisting health problems and/orunusually high levels of stress (yes or no); (13) geographic region (China/Japan, easternEurope,1 western continental Europe2/Israel, the United Kingdom, Scandinavia,3 theUnited States, or Bangladesh/Lebanon); (14) sample size, log transformed; (15)subjective scale of study quality (1–3 range); (16) a continuous composite measureof study quality (0–10 range); (17) a series of variables indicating whether sex, age,socioeconomic status, and health were statistically controlled; and (18) interactionterms between gender, mean age, and follow-up duration.

Results

Table 3 provides descriptive statistics on the 1,377 mortality risk estimates includedin the current meta-analysis. Data were obtained from 123 studies published between1955 and 2007, covering 22 countries, and representing more than 500 million

1 Croatia, Czech Republic, Hungary, Lithuania, and Russia.2 Austria, Belgium, France, Germany, Netherlands, and Spain.3 Denmark, Finland, Norway, and Sweden.

588 E. Shor et al.

Author's personal copy

Table 3 Distribution of mortality risk estimates (n 0 1,377) in the analysis by selected variables

Variable Distribution (%)

Publication Date

1950–1959 4.5

1960–1969 11.4

1970–1979 4.4

1980–1989 14.5

1990–1999 35.5

2000–2008 29.7

Level of Statistical Adjustment

Unadjusted 50.2

Adjusted for age only 20.6

Adjusted for age and additional covariates 29.3

Gender

Women only 45.2

Men only 48.3

Both genders 6.5

Mean Age

<20 0.3

20–29 2.8

30–39 9.4

40–49 19.7

50–59 18.6

60–69 17.4

70–79 22.6

≥80 9.2

Enrollment Start Year

1766–1939 6.6

1940–1949 7.4

1950–1959 8.1

1960–1969 26.4

1970–1979 19.3

1980–1989 23.9

1990–2001 8.3

Comparison Group

Married only 91.5

General population 8.5

Population Consists of Stressed Persons Only

Yes 2.4

No 97.6

Region

Scandinavia 18.6

United States 29.9

United Kingdom 14.1

Eastern Europe 2.7

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 589

Author's personal copy

people. Both men and women are well represented in the data set, and 87.5% of thestudy samples had a mean age of 40 years or older. The median of the maximumfollow-up was five years. Of the HRs analyzed, 91.9% were reported in studiesassigned a subjective quality rating of average or high; the mean five-year impactfactor was 4.2; and the mean number of citations received per year sincepublication was 2.4.

The results of a number of meta-analyses are presented in Table 4 (see Table 5 forsample size information), stratified by the level of statistical adjustment of the riskestimate. They reveal that widowed individuals were more likely to die than theirmarried, nonwidowed counterparts. The mean HR was 1.73 among statisticallyunadjusted point estimates (95% CI, 1.68–1.79; n 0 693 HRs); 1.20 among age-adjusted point estimates (95% CI, 1.15–1.26; n 0 284); and 1.20 among pointestimates adjusted for age and additional covariates (95% CI, 1.16–1.25; n 0 400).Exclusion of HRs based on estimated death rates, and of HRs in which the standarderror was estimated, does not substantively alter the mean HRs (see Table 4). Themean HR among studies with a low subjective quality rating did not differ signifi-cantly from 1.00 (HR, 1.44; 95% CI, 0.90–2.32; n 0 2), but this may be due solely tothe small sample size. The mean HR was elevated among studies with an averagequality rating (HR, 1.17; 95% CI, 1.08–1.26; n 0 104) and the highest quality rating(HR, 1.22; 95% CI, 1.16–1.28; n 0 294). Thus, after controlling for multiplecovariates including age and including only high-quality studies, widowhood wasassociated with a 22% higher risk of mortality.

Subgroup Meta-Analyses and Meta-Regression Analyses

From this point forward, the discussion will focus on the more conservative findingsof HRs adjusted for age and additional covariates (see Table 4). Results of these

Table 3 (continued)

Variable Distribution (%)

Western Continental Europe/Israel 28.6

China/Japan 2.6

Bangladesh/Lebanon 3.5

Follow-up Time

<1.5 years 25

1–5 years 25

5–10 years 25

>10 years 25

Death Rate Estimated?

Yes 23.8

No 76.2

Standard Error Estimated?

Yes 48.4

No 51.6

590 E. Shor et al.

Author's personal copy

Table 4 Meta-analyses of the mortality risk for widows relative to married personsa

UnadjustedAdjusted forAge Only

Adjusted for Age andAdditional Covariatesb

All Available Data 1.73 (1.68, 1.79)*** 1.20 (1.15, 1.26)*** 1.20 (1.16, 1.25)***

Non-estimated Death Rate Only 1.80 (1.74, 1.86)*** 1.28 (1.20, 1.36)*** 1.21 (1.16, 1.27)***

Non-estimated SE Only 1.61 (1.52, 1.71)*** 1.28 (1.23, 1.33)*** 1.23 (1.19, 1.28)***

By Subjective Quality Score

Low 1.59 (1.45, 1.74)*** 0.95 (0.86, 1.04) 1.44 (0.90, 2.32)

Average 1.79 (1.72, 1.85)*** 1.24 (1.15, 1.33)*** 1.17 (1.08, 1.26)***

High 1.61 (1.49, 1.75)*** 1.31 (1.23, 1.41)*** 1.22 (1.16, 1.28)***

By Gender

Women 1.61 (1.54, 1.69)*** 1.05 (0.99, 1.12) 1.15 (1.08, 1.22)***

Men 1.84 (1.76, 1.92)*** 1.39 (1.30, 1.48)*** 1.27 (1.19, 1.35)***

Mean Age

20–29 2.67 (2.16, 3.30)*** –– ––

30–39 2.78 (2.35, 3.30)*** –– 1.95 (0.98, 3.87)

40–49 2.53 (2.34, 2.73)*** 6.23 (3.96, 9.80)*** 1.15 (1.02, 1.29)*

50–59 1.83 (1.71, 1.95)*** 1.63 (1.26, 2.10)*** 1.38 (1.15, 1.67)***

60–69 1.65 (1.54, 1.77)*** 1.32 (1.20, 1.46)*** 1.24 (1.16, 1.34)***

70–79 1.52 (1.41, 1.63)*** 1.16 (1.02, 1.32)* 1.19 (1.07, 1.32)**

≥80 1.46 (1.38, 1.54)*** 1.14 (1.09, 1.20)*** 1.18 (1.11, 1.24)***

Region

Scandinavia 1.43 (1.28, 1.59)*** 1.06 (0.99, 1.13) 1.22 (1.13, 1.32)***

United States 1.60 (1.50, 1.70)*** 1.47 (1.32, 1.65)*** 1.19 (1.12, 1.26)***

United Kingdom 1.32 (1.20, 1.45)*** 1.12 (1.02, 1.23)* 1.16 (1.01, 1.33)*

Eastern Europe 1.96 (1.73, 2.23)*** 1.79 (1.42, 2.25)*** 1.01 (0.57, 1.80)

Western Continental Europe/Israel 1.96 (1.88, 2.05)*** 1.34 (1.22, 1.47)*** 1.25 (1.14, 1.37)***

China/Japan 1.49 (1.11, 2.01)** 1.00 (0.73, 1.38) 1.17 (1.00, 1.37)*

Bangladesh/Lebanon 1.60 (1.38, 1.84)*** … 1.22 (0.97, 1.54)

Enrollment Start Year

1766–1939 0.53 (0.26, 1.07) 1.14 (0.99, 1.31) 1.14 (1.05, 1.24)**

1940–1949 1.61 (1.48, 1.74)*** 1.48 (1.01, 2.16)* ––

1950–1959 1.38 (1.26, 1.50)*** 1.61 (1.31, 2.00)*** ––

1960–1969 1.83 (1.75, 1.93)*** 1.06 (0.98, 1.15) 1.18 (1.03, 1.36)*

1970–1979 1.45 (1.33, 1.57)*** 1.15 (1.06, 1.25)*** 1.17 (1.08, 1.26)***

1980–1989 2.16 (2.03, 2.29)*** 1.26 (1.15, 1.38)*** 1.24 (1.16, 1.33)***

1990–1999 1.34 (1.12, 1.61)** 1.61 (1.39, 1.86)*** 1.27 (1.16, 1.39)***

Follow-up Duration

≤6 months 1.76 (1.55, 1.99)*** 1.48 (1.32, 1.66)*** 1.58 (1.32, 1.88)***

1 year 1.86 (1.75, 1.97)*** 1.43 (1.23, 1.66)*** 1.34 (1.10, 1.62)**

2 years 1.60 (1.48, 1.73)*** 1.33 (1.16, 1.53)*** 1.51 (1.27, 1.79)***

3 years 1.60 (1.47, 1.73)*** 1.08 (0.91, 1.27) 1.20 (0.90, 1.61)

4 years 1.28 (1.09, 1.50)** 1.03 (0.89, 1.20) 1.35 (0.98, 1.86)

5 years 1.30 (1.11, 1.52)*** 0.95 (0.72, 1.25) 1.19 (1.01, 1.41)*

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 591

Author's personal copy

analyses reveal that widowhood had a deleterious effect for both genders, but themagnitude of the effect was greater for men (HR, 1.27; 95% CI, 1.19–1.35; n 0 166)than for women (HR, 1.15; 95% CI, 1.08–1.22; n 0 177). Furthermore, the results ofmeta-regression analyses, modeling all main effects (Model 1), main effects plusthree interaction terms (Model 2), and a final parsimonious model (Model 3), confirmthat the increase in risk of death for men who lost their spouse was substantiallyhigher than the increase in risk for women who lost their spouse (see Table 6).

An interesting result comes from comparing groups by average age at studyenrollment. As shown in Table 4, widowhood has a harmful effect on mortality inalmost all age groups, but the magnitude of the effect decreases with age. The meanHR associated with widowhood was high yet nonsignificant for people aged 30–39(HR, 1.94; 95% CI, 0.98–3.87; n 0 3). The lack of significance, however, is probablydue to the limited number of studies that included individuals in this age range, thesmall number of widows, and the low mortality rate in this age range. It becamesignificant in the 40–49 age group, in which widows had a 15% higher risk ofdeath than married persons (HR, 1.15; 95% CI, 1.02–1.29; n 0 33), andremained so for all other age groups. Although the risk was highest for those aged50–59 (HR, 1.38; 95% CI, 1.15–1.67; n 0 27), it then decreased for those aged 60–69(HR, 1.24; 95% CI, 1.16–1.34; n 0 103), 70–79 (HR, 1.19; 95% CI, 1.07–1.32; n 052), and 80 years or older (HR, 1.18; 95% CI, 1.11–1.24; n 0 182). The results of theinitial meta-regression analysis (Model 1 of Table 6) reflect this downward trendamong the latter four age groups (suggesting a 10% decrease for each additional10 years; p < .001).

Table 4 (continued)

UnadjustedAdjusted forAge Only

Adjusted for Age andAdditional Covariatesb

6 years 1.07 (0.81, 1.43) 1.19 (1.05, 1.34)** 1.15 (1.02, 1.30)*

7 years 1.21 (0.95, 1.53) 1.11 (0.95, 1.29) 1.24 (1.08, 1.41)**

8 years 1.75 (1.47, 2.08)*** 1.25 (0.89, 1.77) 1.11 (0.88, 1.40)

9 years 3.62 (2.73, 4.79)*** 1.01 (0.75, 1.35) 1.19 (1.04, 1.35)**

10 years –– 1.23 (1.11, 1.35)*** 1.18 (1.04, 1.35)*

11 years 1.31 (1.03, 1.68)* 1.10 (0.96, 1.27) 1.25 (0.99, 1.57)

12 years –– 1.20 (0.80, 1.80) 1.52 (1.26, 1.84)***

13 years 1.23 (1.01, 1.49)* 1.22 (0.91, 1.64) 1.13 (0.86, 1.49)

14 years 1.29 (0.98, 1.69) 1.42 (0.95, 2.11) 1.18 (0.77, 1.81)

15 years –– 1.29 (0.75, 2.22) 1.07 (0.90, 1.26)

16–20 years 1.30 (1.09, 1.56)** 0.93 (0.75, 1.15) 1.22 (1.02, 1.47)*

21–25 years –– 1.15 (0.88, 1.49) 1.27 (1.09, 1.49)**

>25 years 1.22 (1.00, 1.48)* 1.03 (0.80, 1.34) 1.11 (1.02, 1.20)*

aAll meta-analyses calculated by maximum likelihood using a random-effects model (n 0 1,381). Numbersshown are mean HRs (95% confidence interval). Dashes indicate instances where n ≤ 1 and meaningfulmean HR could not be calculated. See Table 5 for sample size information.bThe number and type of covariates vary between studies.

*p < .05; **p < .01; ***p < .001; (two-tailed tests)

592 E. Shor et al.

Author's personal copy

Table 5 Heterogeneity test results and sample size information for the meta-analyses reported in Table 4

Unadjusted Adjusted for Age OnlyAdjusted for Age andAdditional Covariates

N p Value FromQ Test

N p Value FromQ Test

N p Value FromQ Test

All Available Data 693 .000 284 .409 400 .999

Non-Estimated DeathRate Only

554 .000 164 .995 331 .999

Non-estimated SE Only 171 .000 200 .814 338 .999

By Subjective QualityScore

Low 99 .643 48 .000 2 .920

Average 448 .000 115 .999 104 .999

High 146 .076 121 .988 294 .999

Gender

Women 307 .015 138 .021 177 .999

Men 360 .000 139 .999 166 .999

Mean Age

20–29.9 20 .000 0 –– 0 ––

30–39.9 25 .344 0 –– 3 .869

40–49.9 82 .000 4 .206 33 .018

50–59.9 110 .920 8 .317 27 .999

60–69.9 135 .119 52 .998 103 .999

70–79.9 117 .670 35 .999 52 .999

≥80 204 .042 185 .007 182 .999

Region

Scandinavia 60 .909 105 .000 92 .999

United States 213 .832 50 .999 146 .999

United Kingdom 111 .403 60 .999 24 .999

Eastern Europe 27 .002 8 .823 2 .633

Western ContinentalEurope/Israel

240 .000 54 .200 101 .999

China/Japan 6 .007 6 .732 24 .946

Bangladesh/Lebanon 36 .839 1 –– 11 .940

Enrollment Start Year

1766–1939 2 .119 21 .999 68 .999

1940–1949 100 .682 3 .988 0 ––

1950–1959 102 .997 9 .707 0 ––

1960–1969 251 .000 77 .000 37 .990

1970–1979 98 .000 85 .999 81 .999

1980–1989 121 .005 61 .999 146 .999

1990–1999 17 .382 24 .000 68 .431

Follow-up Duration

≤6 months 53 .999 40 .947 33 .999

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 593

Author's personal copy

The effects of gender and age on the magnitude of the HR are morecomplex than the meta-analyses with only main effects reveals. Specifically,both Model 2 (full model) and Model 3 (parsimonious model) show a signif-icant interaction effect between these two variables (see Table 6). In Model 3, theexponentiated regression coefficients are 1.75 (95% CI, 1.54–2.00) for gender, 0.93(95% CI, 0.91–0.94) for mean age, and 0.94 (95% CI, 0.92–0.96) for the interactionbetween gender and mean age. Taken together, these results indicate that in middleage, the excess mortality risk associated with widowhood is substantially greater formen than for women, but that this excess risk also declines more rapidly with age formen than it does for women. By age 90, the difference in excess mortality riskbetween men and women is negligible; in fact, the HR for widowhood is approxi-mately 1.0 (no excess risk) for both men and women. Figure 2 shows the predictedhazard rate by age, separately for men and women, based on the estimates fromModel 3 of Table 6 (see Online Resource 1, Section 3, for details).

The results presented in Table 4 also show that the effects of widowhood onmortality remained quite stable throughout the 120 years represented by the studiesthat were sampled for the current analysis. Widowhood had a significant harmfuleffect on mortality in studies with a baseline before 1940 (HR, 1.14; 95% CI, 1.05–1.24; n 0 68), and also in studies with a baseline after 1960. The harmful effect,

Table 5 (continued)

Unadjusted Adjusted for Age OnlyAdjusted for Age andAdditional Covariates

N p Value FromQ Test

N p Value FromQ Test

N p Value FromQ Test

1 year 125 .000 24 .396 30 .999

2 years 77 .088 20 .000 15 .000

3 years 89 .049 13 .000 7 .907

4 years 16 .872 16 .000 7 .934

5 years 18 .631 5 .027 13 .999

6 years 11 .984 25 .695 32 .987

7 years 8 .211 16 .962 24 .999

8 years 23 .727 3 .003 8 .995

9 years 5 .023 6 .111 24 .999

10 years 0 –– 36 .998 36 .999

11 years 7 .922 21 .999 14 .962

12 years 0 –– 2 .553 12 .304

13 years 25 .999 5 .993 6 .998

14 years 9 .702 4 .703 4 .950

15 years 0 –– 2 .794 27 .819

16–20 years 15 .000 19 .988 11 .993

21–25 years 1 –– 5 .991 14 .977

>25 years 17 .011 7 .903 54 .999

594 E. Shor et al.

Author's personal copy

Table 6 Multivariate meta-regression analyses predicting hazard ratio magnitude for widowsa

VariableModel 1: All PredictorsExcept Interaction Terms

Model 2: All PredictorsIncluding Interaction Terms

Model 3:Parsimonious Modelb

Constant 1.27 (1.03, 1.58)* 1.05 (0.85, 1.31) 1.03 (0.87, 1.23)

Proportion of Sample That Is Male 1.21 (1.17, 1.25)*** 1.82 (1.58, 2.10)*** 1.75 (1.54, 2.00)***

Mean Age at Enrollment (decades) 0.90 (0.89, 0.91)*** 0.93 (0.91, 0.94)*** 0.93 (0.91, 0.94)***

Age Range (decades) 0.99 (0.97, 1.00)* 0.98 (0.97, 1.00)* 0.99 (0.98, 1.00)*

Study Age (decades) 0.98 (0.97, 0.99)*** 0.98 (0.97, 0.99)*** 0.98 (0.97, 0.99)***

Enrollment Period (years) 1.01 (1.00, 1.01)*** 1.01 (1.00, 1.01)*** 1.01 (1.00, 1.01)***

Years Between Enrollment andStart of Follow-up

1.06 (1.04, 1.07)*** 1.06 (1.05, 1.07)*** 1.06 (1.05, 1.07)***

Follow-up Duration (decades) 0.98 (0.96, 0.99)*** 0.98 (0.97, 1.00)* 0.98 (0.97, 0.99)***

Log n 1.05 (1.04, 1.06)*** 1.05 (1.04, 1.06)*** 1.05 (1.04, 1.06)***

Publication Age (decades) 0.99 (0.97, 1.01) 0.99 (0.97, 1.01) ––

Interactions

Gender×Mean age at enrollment –– 0.94 (0.91, 0.96)*** 0.94 (0.92, 0.96)***

Gender×Follow-up duration –– 0.99 (0.98, 1.01) ––

Regions

United Kingdom (ref.)

Scandinavia 1.02 (0.95, 1.11) 1.03 (0.95, 1.11) 1.01 (0.94, 1.08)

United States 1.17 (1.08, 1.27)*** 1.17 (1.08, 1.28)*** 1.14 (1.06, 1.22)***

Eastern Europe 1.17 (1.05, 1.30)** 1.17 (1.06, 1.30)** 1.18 (1.07, 1.30)**

Western Continental Europe/Israel 1.16 (1.08, 1.25)*** 1.16 (1.08, 1.25)*** 1.13 (1.06, 1.21)***

China/Japan 1.03 (0.90, 1.17) 1.04 (0.91, 1.18) 1.02 (0.90, 1.16)

Bangladesh/Lebanon 1.59 (1.38, 1.82)*** 1.60 (1.40, 1.83)*** 1.57 (1.38, 1.78)***

Controls (1 0 yes)

Gender 0.97 (0.90, 1.05) 0.97 (0.90, 1.05) ––

Age 0.85 (0.81, 0.89)*** 0.85 (0.81, 0.89)*** 0.86 (0.82, 0.91)***

Other demographics 0.99 (0.92, 1.06) 0.98 (0.92, 1.06) ––

SES 0.88 (0.81, 0.95)*** 0.88 (0.81, 0.95)*** 0.89 (0.83, 0.95)***

Health 1.03 (0.95, 1.12) 1.03 (0.95, 1.11) ––

Social ties 1.10 (1.02, 1.18)* 1.10 (1.02, 1.18)* 1.09 (1.02, 1.18)*

Previous stress 0.91 (0.83, 0.99)* 0.91 (0.83, 0.99)* 0.91 (0.84, 0.99)*

Comparison Group is GeneralPopulation (1 0 yes)

1.07 (0.95, 1.20) 1.08 (0.96, 1.21) ––

Stressed Population (1 0 yes) 0.93 (0.81, 1.08) 0.95 (0.82, 1.09) ––

Standard Error Imputed (1 0 yes) 0.97 (0.92, 1.03) 0.98 (0.92, 1.04) ––

Death Rate Imputed (1 0 yes) 0.89 (0.85, 0.93)*** 0.89 (0.85, 0.93)*** 0.88 (0.84, 0.92)***

Subjective Quality Assessment 1.13 (1.08, 1.18)*** 1.13 (1.08, 1.18)*** 1.14 (1.10, 1.18)***

Scale Measure of Study Quality 1.01 (0.99, 1.03) 1.01 (0.99, 1.03) ––

R2 .578 .594 .590

Variance Component 0.0480*** 0.0446*** 0.0453***

aAll meta-regressions calculated by maximum likelihood using a random-effects model (n 0 1,377 for allmodels). Numbers reported are the exponentiated regression coefficient (95% confidence intervals are inparentheses). Dashes indicate instances when a variable was not included in the model.bObtained using backward elimination; variable removed if p > .10.

*p < .05; **p < .01; ***p < .001 (two-tailed tests)

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 595

Author's personal copy

however, was lower in the 1960s and 1970s than in more recent decades. The meanHR was 1.18 (95% CI, 1.03–1.36; n 0 37) in studies with a baseline between 1960and 1969, and was 1.17 (95% CI, 1.08–1.26; n 0 81) in studies with a baselinebetween 1970 and 1979. It increased to 1.24 (95% CI, 1.16–1.33; n 0 146) in studieswith a baseline between 1980 and 1989, and to 1.27 (95% CI, 1.16–1.24; n 0 68) instudies with a baseline between 1990 and 1999. The meta-regression results (Table 6)confirmed that the effect of widowhood on mortality was lower in previous decades.HRs were 2% lower for each additional 10 years that had elapsed since the baselinedata were collected (p < .001).

Follow-up duration was also a significant predictor in the meta-regression analyses(see Table 6), and the meta-analyses suggest that the effects of widowhood onmortality are substantively higher during the first two years of follow-up (see Table 4).The excess risk associated with widowhood was 58% in studies with only six monthsof follow-up (HR, 1.58; 95% CI, 1.32–1.88; n 0 33), 33% in studies with one year offollow-up (HR, 1.33; 95% CI, 1.11–1.61; n 0 30), and 51% in studies with two yearsof follow-up (HR, 1.51; 95% CI, 1.27–1.79; n 0 15). In studies that followedindividuals for 16–20 years, the excess risk decreases to 22% (HR, 1.22; 95% CI,1.02–1.47; n 0 11), 27% for 21–25 years of follow-up (HR, 1.27; 95% CI, 1.09–1.49;n 0 14), and 11% for 25 years or more of follow-up (HR, 1.11; 95% CI, 1.02–1.20;n 0 54). The final regression (Model 3 of Table 6) indicates that the mean HRdecreases by 2% (p < .001) for every additional 10 years of follow-up. This pattern ofresults suggests that the excess risk associated with widowhood is greatest during thefirst few years after the death of a spouse, but persists at reduced levels for 20 years ormore. The hypothesized interaction between follow-up and gender, however, is notsupported (p 0 .244; Model 2 of Table 6).

Finally, the results presented in Table 4 show that the effect of widowhood onmortality is relatively homogenous in different regions of the world. The mean HRwas 1.22 for Scandinavia (95% CI, 1.13–1.32; n 0 92); 1.19 for the United States(95% CI, 1.12–1.26; n 0 146); 1.16 for the United Kingdom (95% CI, 1.01–1.33; n 024); 1.01 for eastern Europe (95% CI, 0.57–1.80; n 0 2); 1.25 for western continentalEurope (95% CI, 1.14–1.36; n 0 101); 1.17 for China and Japan (95% CI, 1.00–1.37;n 0 24); and 1.22 for Bangladesh/Lebanon (95% CI, 0.97–1.54; n 0 11). Model 3 inTable 6 suggests that in the United States, eastern Europe, western Europe, andBangladesh/Lebanon, widowed people have a somewhat higher risk for mortality

0.40

0.60

0.80

1.00

1.20

1.40

1.60

1.80

2.00

40 50 60 70 80 90M

ean

HR

Mean Age (years)

Men

Women

Fig. 2 Mean hazard ratio bymean age and gender, based onModel 3 of Table 6

596 E. Shor et al.

Author's personal copy

than in the United Kingdom (the reference group), in Scandinavia (p 0 .844), and inChina and Japan (p 0 .716). This model shows that the magnitude of the effect is 14%higher in the United States (p < .001), 18% higher in eastern Europe (p 0 .001), and13% higher in western Europe (p < .001). Although Model 3 shows that the mean HRis 57% higher in Bangladesh/Lebanon (p < .001), this is likely due solely to the factthat there are more than three times as many unadjusted HRs (n 0 36) as there areHRs adjusted for age and additional covariates (n 0 11) for this region.

The results presented in Table 6 show that other significant predictors of differ-ences among reported HRs include the time elapsed between a study’s end ofparticipant enrollment and beginning of follow-up (a 6% increase in risk for eachadditional year; p < .001), and whether the risk estimate was adjusted for age (a 14%decrease when age was controlled; p < .001), SES (a 11% decrease when controlled;p < .001), social ties (a 9% increase when controlled; p 0 .013), and previous stress (a9% decrease when controlled; p 0 .034). The results presented in Table 6 also showthat HRs in studies that estimated the underlying death rate were significantly lowerthan in studies that did not (a 12% decrease; p < .001). Finally, contrary to thecommon conception that the average effect size decreases as the study qualityimproves, we found that the mean magnitude of the effect actually increased instudies that were evaluated as having a higher quality (a 14% increase in the hazardfor each 1-point increase in the 3-point subjective study-quality measure; p < .001).

Analysis of Data Heterogeneity

The between-groups Cochrane’s Q for the meta-analysis of all 1,377 HRs wasstatistically significant (p < .001), and the I2 statistic was quite high (I2, 99.2; 95% CI,98.8–99.5), indicating that important moderating variables exist and supporting thedecisions to use random-effects models and conduct subgroup meta-analyses. Be-cause the discussion of the meta-analysis focused on HRs adjusted for age andadditional covariates, the corresponding heterogeneity test results were carefullyexamined. As shown in Table 5, the Q tests for these subgroup meta-analyses werestatistically significant for only two cases: the 40–49 age group (p 0 .018) and thetwo-year follow-up group (p < .001). I2 tests for these subgroups indicateheterogeneity was moderate for the 40–49 age group (I2, 37.2; 95% CI, 4.2–58.8)and high for the two-year follow-up group (I2, 67.0; 95% CI, 43.4–80.8). The resultsfrom these two subgroup meta-analyses should therefore be treated conservative-ly. In all remaining subgroup analyses, however, Q tests and I2 tests werenonsignificant, indicating that heterogeneity was adequately accounted for by theuse of a random effects model.

Meta-regressions were also used to examine possible sources of heterogeneity inthe data. The model fit statistics for Model 3 of Table 6 (R2, .590; p < .001 for theCochrane’s Q of the model) indicate that this model captured a very substantialportion of the heterogeneity in the data. Nevertheless, the unexplainedheterogeneity variance component (which measures the nonrandom varianceremaining in the model residuals after the effects of all independent variables havebeen taken into account) for the models shown in Table 6 remained highly significant(each p < .001), confirming the need to use a random effects model for all analyses.

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 597

Author's personal copy

Discussion

The results of the present meta-analyses and meta-regression analyses show that,overall, the relative risk of death for those who lost their spouse was 22% higher thanthe risk among married persons, among high-quality studies that adjusted for age andadditional covariates. The adverse effects of widowhood on mortality, however, werenot uniform across all subgroups. As hypothesized, the effects were greater for men(an average increased risk of 27%) than for women (an average increased risk of only15%), with these risks and the difference between them being more pronounced atyounger ages and less pronounced at older ages. By age 90, no difference was foundbetween widowed and married persons among men or women. This aspect of thefindings is consistent with Manzoli et al. (2007), who also found no difference inrelative risk between men and women at older ages. They are also consistent withMineau et al. (2002) and Smith and Zick (1996), who found that the relative mortalityrisk was higher for widowed men than for widowed women and that the relative riskwas higher at younger ages than at older ages. Because these previous studiesexamined only the independent effects of gender and age, however, the documenta-tion of an interaction between gender and age in the present study is a major finding.

A comparison between findings from earlier and more recent studies revealed thatthe excess risk of mortality among widowed persons has been slowly increasing overtime. This both supports the hypothesis put forth earlier in this article and suggeststhat future meta-analyses should strive to include the results from both early andrecent studies in order to evaluate the impact of societal trends. The role of marriagein networks of interpersonal ties has shifted over time (Henrard 1996; Manzoli et al.2007), and multiple facets of this shift may be reflected in the time trend of increasingHRs. In previous decades, widowed men almost always remarried. Because widowedwomen have always outnumbered widowed men, the long-term widowed group waspredominately female. Declining rates of remarriage in Western nations (Bramlett andMosher 2002; Bumpass and Sweet 1991) have increased the relative number of menwho are in the long-term widowed group in more recent years. Because widowedmen have a higher relative risk than do widowed women, the growing proportion ofmale widowers would cause the overall HR to rise over time. In addition, rates ofcohabitation have increased. Research has shown that, controlling for age, those whochoose cohabitation tend to have lower SES and therefore are likely to be less healthythan those who marry (Manning and Smock 2002). Presuming that those whocohabitate would have married in previous decades, the growth of the less-healthycohabitating group increased the average health level of the denominator (married)group over time. This also would cause the overall HR to rise. Factors that help bufferthe stress of widowhood have also become less available. Societal decentralizationand the geographic dispersal of the family have altered the quantity and quality ofinterpersonal social support available to widows (Lopata 1978; Popenoe 1993). Theerosion of pensions and other similar supplemental sources of income since the 1970shas brought new challenges for widows in maintaining their pre-widowhood materialquality of life (Marin and Zolyomi 2010). The loss of buffers like this would also helpexplain rising HRs over time. Finally, the married population has benefited most fromcertain health care advances, such as the prevention of childbearing-related deaths.Likewise, compared with widowed persons, married persons have benefitted much

598 E. Shor et al.

Author's personal copy

more from family-oriented primary care strategies (McDaniel et al. 2005). Positivehealth changes such as these, which bring about a reduction in the mortality rate forthe denominator population, can also explain the increase in the HR over time.

An interesting finding emerges from the current analyses concerning the differencein the effect of widowhood on mortality by the duration of follow-up, which exhibitsthe pattern hypothesized by Young et al. (1963). Our findings add to the literature(e.g., Nystedt 2002; Thierry 1999) on this topic by providing consensus estimates forthe length of the period immediately following widowhood when the survivingspouse is at his/her greatest risk. As seen in Table 4, the risk of mortality is especiallyhigh in studies that followed individuals for two years or less. The excess riskdecreases substantially among studies with longer follow-up durations, although itremains elevated among studies with up to 15 years of follow-up. Although the onsetof widowhood coincided with the enrollment period for only a small fraction of thestudies, these findings suggest that the immediate stress caused by widowhood isindeed an important factor in increasing the risk of mortality. Further analysesshould investigate the specific physiological and/or behavioral mechanisms thatlead to the increase of risk during the first two years after losing a spouse. Theresults suggest the possibility that different mechanisms dominate the early and laterstages of widowhood. Comorbidity effects (Cheung 2000; Elwert and Christakis2008; Lillard and Panis 1996; Smith and Zick 1996) may combine with stress effectsin the early years of widowhood but decline as the influence of the lost spousediminishes. Practitioners and counselors should focus their attention on the first yearsof widowhood without losing sight of the continuing risk. Identification of theunderlying pathophysiology and determining the changing contributions of physio-logical and behavioral mechanisms over time will contribute to better targeting ofsupportive interventions.

Finally, the analysis by region of the world suggests that the risk of deathfollowing widowhood is approximately equal in most regions. The magnitude ofthe effect in the less-developed countries—eastern European countries as well asBangladesh/Lebanon—is of particular interest. Economic support may have in-creased importance in poorer countries, where the decrease in income associatedwith the loss of a spouse may substantially reduce the quality of nutrition andhealthcare. We cannot, however, make firm conclusions about the mean effect indeveloping nations because of the small number of studies conducted in them.Although the mean HR is suggestively high in eastern Europe among HRs adjustedfor age alone, there are not enough studies to evaluate whether this pattern would holdamong HRs adjusted for age and additional covariates. The results for Bangladesh/Lebanon should be treated with caution as well, considering the small number ofstudies conducted.

Limitations

A major limitation of the reported analyses, shared by many meta-analyses, is the“file drawer effect,” or more specifically, the nonreporting in the literature of non-significant findings (Berman and Parker 2002; Egger and Davey-Smith 1998). Thistendency may lead to an overestimation of the mean HRs. Therefore, one should beespecially careful in interpreting mean HRs that are relatively close to 1, even when

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 599

Author's personal copy

these are significant (as is the case with some of the results in the current meta-analysis). A funnel plot of the log HRs against sample size appears somewhatasymmetric around the mean HR, suggesting the possibility of publication bias(Fig. 3). The results of formal tests for publication bias differ, with Eggers’ test(Egger and Davey-Smith 1998) indicating significant bias (p < .001) and that ofPeters et al. (Moreno et al. 2009; Peters et al. 2006) indicating no significant bias(p 0 .178). The results of the more conservative Eggers test suggest that the HRs thatare missing from the analysis are small studies with large HRs. The nature of the biasis such that our results would tend to underestimate the mean HR rather thanoverestimate it.

A second limitation stems from the nature of the data. Most of the research onwidowhood and mortality was conducted in the developed world. Very few studieswere conducted in eastern European countries, the Middle East, or South Asia,and there were none from Africa or South America. The sample sizes in thestudies from the developing world are small, and conclusions cannot be drawnabout potential differences between the developed and the developing world. Also,because most of the results come from the developed countries, the findings fromthe different analyses presented here should not be extrapolated to populations indeveloping countries. In addition, important moderators—such as race, ethnicity,and occupational class—were not examined because of data unavailability. Futurestudies, stratified by these factors or including appropriate interaction terms, areneeded.

Conclusion

In conclusion, the analyses reported here show that widowhood substantiallyincreases the risk of death among broad segments of the population. Future researchshould focus on understanding the health, socioeconomic, physiological, and behav-ioral factors through which this effect is manifested, especially for younger men andduring the first two years following the loss of a spouse. In addition, results from thefew studies that were conducted in the developing world suggest that widowedpeople in these countries may be at greater risk. Further research in developing

0

200,000

400,000

600,000

800,000

1,000,000

1,200,000

–1.50 –1.00 –0.50 0.00 0.50 1.00 1.50 2.00 2.50

Sam

ple

Size

Hazard Ratio (logged)

Fig. 3 Funnel plot of hazard ra-tios (logged) versus sample size:Hazard ratios statistically adjust-ed for age and additional covari-ates. Vertical line denotes themean hazard ratio (logged) of0.1866

600 E. Shor et al.

Author's personal copy

countries may help explain not only the cultural differences in the experience ofwidowhood but also the differential mechanisms that mediate the risk of deathfollowing widowhood.

Acknowledgments The authors are grateful for the support provided by Grant HL-76857 from theNational Institutes of Health. The funding source had no involvement in the collection, analysis,and interpretation of the data; in the writing of the report; and in the decision to submit the articlefor publication.

References

Alter, G., Dribe, M., & van Poppel, F. W. A. (2007). Widowhood, family size, and post-reproductivemortality: A comparative analysis of three populations in nineteenth-century Europe. Demography, 44,785–806.

Armenian, H. K., Saadeh, F. M., & Armenian, S. L. (1987). Widowhood and mortality in an Armenianchurch parish in Lebanon. American Journal of Epidemiology, 125, 127–132.

Bagiella, E., Hong, V., & Sloan, R. P. (2005). Religious attendance as a predictor of survival in the EPESEcohorts. International Journal of Epidemiology, 34, 443–451.

Ben-Shlomo, Y., Smith, G. D., Shipley, M., & Marmot, M. G. (1993). Magnitude and causes of mortalitydifferences between married and unmarried men. Journal of Epidemiology and Community Health, 47,200–205.

Berkson, J. (1962). Mortality and marital status: Reflections on the derivation of etiology from statistics.American Journal of Epidemiology, 52, 1318–1329.

Berman, N. G., & Parker, R. A. (2002). Meta-analysis: Neither quick nor easy. BMC Medical ResearchMethodology, 2, 10–18.

Bowling, A., & Benjamin, B. (1985). Mortality after bereavement: A follow-up study of a sample of elderlywidowed people. Biology and Society, 2, 197–203.

Bowling, A., & Charlton, J. (1987). Risk factors for mortality after bereavement: A logistic regressionanalysis. The Journal of the Royal College of General Practitioners, 37, 551–554.

Bramlett, M. D., & Mosher, W. D. (2002). Cohabitation, divorce, and remarriage in the United States.National Center for Health Statistics: Vital and Health Statistics, 23(22).

Breeze, E., Sloggett, A., & Fletcher, A. (1999). Socioeconomic and demographic predictors of mortalityand institutional residence among middle-aged and older people: Results from the longitudinal study.Journal of Epidemiology and Community Health, 53, 765–774.

Brockmann, H., & Klein, T. (2002). Family biography and mortality in East and West Germany (inGerman). Zeitschrift Fur Gerontologie Und Geriatrie, 35, 430–440.

Brockmann, H., & Klein, T. (2004). Love and death in Germany: The marital biography and its effect onmortality. Journal of Marriage and Family, 66, 567–581.

Buckley, T., McKinley, S., Tofler, G., & Bartrop, R. (2010). Cardiovascular risk in early bereavement: Aliterature review and proposed mechanisms. International Journal of Nursing Studies, 47, 229–238.

Bumpass, L. L., & Sweet, J. A. (1991). The role of cohabitation in declining rates of marriage. Journal ofMarriage and the Family, 53, 913–927.

Burgoa, M., Regidor, E., Rodriguez, C., & Gutierrez-Fisac, J. L. (1998). Mortality by cause of death andmarital status in Spain. European Journal of Public Health, 8, 37–42.

Campbell, C., & Lee, J. Z. (1996). A death in the family: Household structure and mortality in ruralLiaoning: Life-event and time-series analysis, 1792–1867. The History of the Family, 1, 297–328.

Cheung, Y. B. (2000). Marital status and mortality in British women: A longitudinal study. InternationalJournal of Epidemiology, 29, 93–99.

Christakis, N. A., & Allison, P. (2006). Mortality after the hospitalization of a spouse. The New EnglandJournal of Medicine, 354, 719–730.

Clayton, P. J. (1974). Mortality and morbidity in first year of widowhood. Archives of General Psychiatry,30, 747–750.

Comstock, G. W., & Tonascia, J. A. (1977). Education and mortality in Washington County, Maryland.Journal of Health and Social Behavior, 18, 54–61.

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 601

Author's personal copy

Doblhammer, G. (2000). Reproductive history and mortality later in life: A comparative study in Englandand Wales and Austria. Population Studies, 54, 169–176.

Durkheim, E. (1951). Suicide. Glencoe, IL: Free Press. (Original work published 1897)Dzurova, D. (2000). Mortality differentials in the Czech Republic during the post-1989 socio-political

transformation. Health & Place, 6, 351–362.Ebrahim, S., Wannamethee, G., McCallum, A., Walker, M., & Shaper, A. G. (1995). Marital status, change

in marital status, and mortality in middle-aged British men. American Journal of Epidemiology, 142,834–842.

Egger, M., & Davey-Smith, G. (1998). Meta-analysis: Bias in location and selection of studies. BritishMedical Journal, 316, 61–66.

Ekblom, B. (1963). Significance of socio-psychological factors with regard to risk of death among elderlypersons. Acta Psychiatrica Scandinavica, 39, 627–633.

Elwert, F., & Christakis, N. A. (2008). The effect of widowhood on mortality by the causes of death of bothspouses. American Journal of Public Health, 98, 2092–2098.

Espinosa, J., & Evans, W. N. (2008). Heightened mortality after the death of a spouse: Marriage protectionor marriage selection? Journal of Health Economics, 27, 1326–1342.

Gerra, G., Monti, D., Panerai, A. E., Sacerdote, P., Anderlini, R., Avanzini, P., . . . Franceschi, C. (2003).Long-term immune-endocrine effects of bereavement: Relationships with anxiety levels and mood.Psychiatry Research, 121, 145–158.

Goforth, H. W., Lowery, J., Cutson, T. M., McMillan, E. S., Kenedi, C., & Cohen, M. A. (2009). Impact ofbereavement on progression of AIDS and HIV infection: A review. Psychosomatics, 50, 433–439.

Goldman, N., & Hu, Y. R. (1993). Excess mortality among the unmarried: A case study of Japan. SocialScience & Medicine, 36, 533–546.

Goldman, N., Korenman, S., & Weinstein, R. (1995). Marital status and health among the elderly. SocialScience & Medicine, 40, 1717–1730.

Goodwin, J. S., Hunt, W. C., Key, C. R., & Samet, J. M. (1987). The effect of marital status on stage,treatment, and survival of cancer patients. Journal of the American Medical Association, 258, 3125–3130.

Gove, W. R. (1973). Sex, marital status, and mortality. The American Journal of Sociology, 79, 45–67.Grundy, E., & Kravdal, O. (2008). Reproductive history and mortality in late middle age among Norwegian

men and women. American Journal of Epidemiology, 167, 271–279.Hajdu, P., McKee, M., & Bojan, F. (1995). Changes in premature mortality differentials by marital status in

Hungary and in England and Wales. European Journal of Public Health, 5, 259–264.Hart, C. L., Hole, D. J., Lawlor, D. A., Smith, G. D., & Lever, T. F. (2007). Effect of conjugal bereavement

on mortality of the bereaved spouse in participants of the Renfrew/Paisley Study. Journal of Epide-miology and Community Health, 61, 455–460.

Hayward, M. D., & Gorman, B. K. (2004). The long arm of childhood: The influence of early-life socialconditions on men’s mortality. Demography, 41, 87–107.

Helsing, K. J., & Szklo, M. (1981). Mortality after bereavement. American Journal of Epidemiology, 114,41–52.

Helsing, K. J., Szklo, M., & Comstock, G. W. (1981). Factors associated with mortality after widowhood.American Journal of Public Health, 71, 802–809.

Helweg-Larsen, M., Kjoller, M., & Thoning, H. (2003). Do age and social relations moderate therelationship between self-rated health and mortality among adult Danes? Social Science & Medicine,57, 1237–1247.

Hemstrom, O. (1996). Is marriage dissolution linked to differences in mortality risks for men and women?Journal of Marriage and the Family, 58, 366–378.

Henrard, J. C. (1996). Cultural problems of aging especially regarding gender and intergenerational equity.Social Science & Medicine, 43, 667–680.

Henretta, J. C. (2007). Early childbearing, marital status, and women’s health and mortality after age 50.Journal of Health and Social Behavior, 48, 254–266.

Horwitz, O., & Weber, J. (1974). Marital status and mortality (in Danish). Sygeplejersken, 74, 10–14.Huedo-Medina, T. B., Sanchez-Meca, J., & Marin-Martinez, F. (2006). Assessing heterogeneity in meta-

analysis: Q statistic or I2 index? Psychological Methods, 11, 193–206.Hughes, M. E., & Waite, L. J. (2009). Marital biography and health at mid-life. Journal of Health and

Social Behavior, 50, 344–358.Ikeda, A., Iso, H., Toyoshima, H., Fujino, Y., Mizoue, T., Yoshimura, T., . . . JACC Study Group. (2007).

Marital status and mortality among Japanese men and women: The Japan Collaborative Cohort Study.BMC Public Health, 7:73. doi:10.1186/1471-2458-7-73

602 E. Shor et al.

Author's personal copy

Iribarren, C., Jacobs, D. R., Kiefe, C. I., Lewis, C. E., Matthews, K. A., Roseman, J. M., & Hulley, S. B.(2005). Causes and demographic, medical, lifestyle and psychosocial predictors of premature mortal-ity: The CARDIA Study. Social Science & Medicine, 60, 471–482.

ISI Web of Knowledge. (2009). Journal citation reports. New York: Thomson Reuters. Retrieved fromhttp://wokinfo.com/products_tools/analytical/jcr/

Iwasaki, M., Otani, T., Sunaga, R., Miyazaki, H., Xiao, L., Wang, N., . . . Suzuki, S. (2002). Socialnetworks and mortality based on the Komo-Ise Cohort Study in Japan. International Journal ofEpidemiology, 31, 1208–1218.

Jagger, C., & Sutton, C. J. (1991). Death after marital bereavement—Is the risk increased? Statistics inMedicine, 10, 395–404.

Jenkinson, C. M., Madeley, R. J., Mitchell, J. R. A., & Turner, I. D. (1993). The influence of psychosocialfactors on survival after myocardial infarction. Public Health, 107, 305–317.

Jin, L., & Christakis, N. A. (2009). Investigating the mechanism of marital mortality reduction: Thetransition to widowhood and quality of health care. Demography, 46, 605–625.

Johansen, C., Schou, G., Soll-Johanning, H., Mellemgaard, A., & Lynge, E. (1996). Influence of maritalstatus on survival from colon and rectal cancer in Denmark. British Journal of Cancer, 74, 985–988.

Johnson, N. J., Backlund, E., Sorlie, P. D., & Loveless, C. A. (2000). Marital status and mortality: TheNational Longitudinal Mortality Study. Annals of Epidemiology, 10, 224–238.

Jones, D. R., & Goldblatt, P. O. (1987). Cause of death in widow(er)s and spouses. Journal of BiosocialScience, 19, 107–121.

Jones, D. R., & Goldblatt, P. O. (2006). Cancer mortality following widow(er)hood: Some further resultsfrom the Office of Population Censuses and Surveys Longitudinal Study. Stress and Health, 2, 129–140.

Jones, D. R., Goldblatt, P. O., & Leon, D. A. (1984). Bereavement and cancer: Some data on death ofspouses from the Longitudinal Study of Office of Population Censuses and Surveys. British MedicalJournal, 289, 461–464.

Joung, I. M. A., Glerum, J. J., van Poppel, J. W. P. F., & Mackenbach, J. P. (1996). The contribution ofspecific causes of death to mortality differences by marital status in the Netherlands. European Journalof Public Health, 6, 142–149.

Jylha, M., & Aro, S. (1989). Social ties and survival among the elderly in Tampere, Finland. InternationalJournal of Epidemiology, 18, 158–164.

Kalediene, R., Petrauskiene, J., & Starkuviene, S. (2007). Inequalities in mortality by marital status duringsocio-economic transition in Lithuania. Public Health, 121, 385–392.

Kaplan, R. M., & Kronick, R. G. (2006). Marital status and longevity in the United States population.Journal of Epidemiology and Community Health, 60, 760–765.

Kaprio, J., Koskenvuo, M., & Rita, H. (1987). Mortality after bereavement—A prospective study of 95,647widowed persons. American Journal of Public Health, 77, 283–287.

Keller, A. Z. (1969). Survivorship with mouth and pharynx cancer and their association with cirrhosis of theliver, marital status, and residence. American Journal of Public Health Nations Health, 59, 1139–1153.

Kohler, I., & Kohler, H.-P. (2002). The effect of marriage on mortality: Selection or protection? Evidencefrom Danish Twins using fixed-effect survival models. Unpublished document, Population StudiesCenter, University of Pennsylvania.

Kolip, P. (2005). The association between gender, family status and mortality. Journal of Public Health, 13,309–312.

Kraus, A., & Lilienfeld, A. (1959). Some epidemiological aspects of the high mortality rate in the youngwidowed group. Journal of Chronic Diseases, 10, 207–217.

Kravdal, O. (2003). Children, family and cancer survival in Norway. International Journal of Cancer, 105,261–266.

Kravdal, O. (2007). A fixed-effects multilevel analysis of how community family structure affectsindividual mortality in Norway. Demography, 44, 519–537.

Kroenke, C. H., Kubzansky, L. D., Schernhammer, E. S., Holmes, M. D., & Kawachi, I. (2006). Socialnetworks, social support, and survival after breast cancer diagnosis. Journal of Clinical Oncology, 24,1105–1111.

Kumle, M., & Lund, E. (2000). Patterns of childbearing and mortality in Norwegian women. A 20-yearfollow-up of women aged 40–96 in the 1970 Norwegian census. In J.-M. Robine, T. B. L. Kirkwood,& M. Allard (Eds.), Sex and longevity: Sexuality, gender, reproduction, parenthood (pp. 117–128).Berlin, Germany: Springer.

Lichtenstein, P., Gatz, M., & Berg, S. (1998). A twin study of mortality after spousal bereavement.Psychological Medicine, 28, 635–643.

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 603

Author's personal copy

Lillard, L. A., & Panis, C. W. A. (1996). Marital status and mortality: The role of health. Demography, 33,313–327.

Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, CA: Sage.Liu, L. J., & Sullivan, D. H. (2003). Relationship between social demographic factors and survival within

one year of hospital discharge in a cohort of elderly male patients. Journal of Epidemiology, 13, 203–210.

Lopata, H. Z. (1978). Contributions of extended families to the support systems of metropolitanarea widows: Limitations of the modified kin network. Journal of Marriage and the Family, 40,355–364.

Lusyne, P., Page, H., & Lievens, J. (2001). Mortality following conjugal bereavement, Belgium 1991–96:The unexpected effect of education. Population Studies, 55, 281–289.

Makuc, D. M., Feldman, J. J., Kleinman, J. C., & Pierre, M. B. J. (1990). Sociodemographic differentials inmortality. In J. Coroni-Huntley, R. R. Huntley, & J. J. Feldman (Eds.), Health status and well-being ofthe elderly: National Health and Nutrition Examination Survey-I Epidemiologic Follow-up Study (pp.155–171). New York: Oxford University Press.

Malyutina, S., Bobak, M., Simonova, G., Gafarov, V., Nikitin, Y., & Marmot, M. (2004). Education, maritalstatus, and total and cardiovascular mortality in Novosibirsk, Russia: A prospective cohort study.Annals of Epidemiology, 14, 244–249.

Manning, W. D., & Smock, P. J. (2002). First comes cohabitation and then comes marriage? A researchnote. Journal of Family Issues, 23, 1065–1087.

Manor, O., & Eisenbach, Z. (2003). Mortality after spousal loss: Are there socio-demographic differences?Social Science & Medicine, 56, 405–413.

Manor, O., Eisenbach, Z., Peritz, E., & Friedlander, Y. (1999). Mortality differentials among Israeli men.American Journal of Public Health, 89, 1807–1813.

Manor, O., Eisenbach, Z., Peritz, E., & Friedlander, Y. (2000). Mortality differentials among women: TheIsrael Longitudinal Mortality Study. Social Science & Medicine, 51, 1175–1188.

Manzoli, L., Villari, P., Pirone, G. M., & Boccia, A. (2007). Marital status and mortality in the elderly: Asystematic review and meta-analysis. Social Science & Medicine, 64, 77–94.

Marin, B., & Zolyomi, E. (2010). Women’s work and pensions: What is good, what is best? Designinggender-sensitive arrangements. Vienna, Austria: Ashgate Publishing.

Martelin, T. (1996). Socio-demographic differentials in mortality at older ages in Finland. In G. Caselli &A. D. Lopez (Eds.), Health and mortality among elderly population (pp. 112–134). New York:Clarendon Press.

Martelin, T., Koskinen, S., & Valkonen, T. (1998). Sociodemographic mortality differences among theoldest old in Finland. Journals of Gerontology Series B-Psychological Sciences and Social Sciences,53, S83–S90.

Martikainen, P. T. (1990). Unemployment and mortality among Finnish men, 1981–5. British MedicalJournal, 301, 407–411.

Martikainen, P. T. (1995). Women’s employment, marriage, motherhood and mortality: A test of themultiple role and role accumulation hypotheses. Social Science & Medicine, 40, 199–212.

Martikainen, P. T., Martelin, T., Nihtila, E., Majamaa, K., & Koskinen, S. (2005). Differences inmortality by marital status in Finland from 1976 to 2000: Analyses of changes in marital-statusdistribution, socio-demographic and household composition, and cause of death. PopulationStudies, 59, 99–115.

Martikainen, P. T., & Valkonen, T. (1996a). Mortality after death of spouse in relation to duration ofbereavement in Finland. Journal of Epidemiology and Community Health, 50, 264–268.

Martikainen, P. T., & Valkonen, T. (1996b). Mortality after the death of a spouse: Rates and causes of deathin a large Finnish cohort. American Journal of Public Health, 86, 1087–1093.

Martikainen, P. T., & Valkonen, T. (1998). Do education and income buffer the effects of death of spouse onmortality? Epidemiology, 9, 530–534.

McDaniel, S. H., Campbell, T. L., Hepworth, J., & Lorenz, A. (2005). Family-oriented primary care (2nded.). New York: Springer.

Mellstrom, D., Nilsson, A., Oden, A., Rundgren, A., & Svanborg, A. (1982). Mortality among the widowedin Sweden. Scandinavian Journal of Social Medicine, 10, 33–41.

Mendes De Leon, C. F., Appels, W. P. M. A., Otten, F. W. J., & Schouten, E. G. W. (1992). Risk ofmortality and coronary heart disease by marital status in middle-aged men in the Netherlands.International Journal of Epidemiology, 21, 460–466.

Mendes De Leon, C. F., Kasl, S. V., & Jacobs, S. (1993). Widowhood and mortality risk in a communitysample of the elderly: A prospective study. Journal of Clinical Epidemiology, 46, 519–527.

604 E. Shor et al.

Author's personal copy

Metayer, C., Coughlin, S. S., & McCarthy, E. P. (1996). Marital status as a predictor of survival inidiopathic dilated cardiomyopathy: The Washington, DC, Dilated Cardiomyopathy Study. EuropeanJournal of Epidemiology, 12, 573–582.

Mineau, G. P., Smith, K. R., & Bean, L. L. (2002). Historical trends of survival among widows andwidowers. Social Science & Medicine, 54, 245–254.

Mollica, R. F., Srajlic, N., Chernoff, M., Lavelle, J., Vukovic, I. S., & Massagli, M. P. (2001). Longitudinalstudy of psychiatric symptoms, disability, mortality, and emigration among Bosnian refugees. Journalof the American Medical Association, 286, 546–554.

Moreno, S. G., Sutton, A. J., Ades, A. E., Stanley, T. D., Abrams, K. R., Peters, J. L., & Cooper, N. J.(2009). Assessment of regression-based methods to adjust for publication bias through a comprehen-sive study. BMC Medical Research Methodology, 9, 2. doi:10.1186/1471-2288-9-2

Mostafa, G., & van Ginneken, J. K. (2000). Trends in and determinants of mortality in the elderlypopulation of Matlab, Bangladesh. Social Science & Medicine, 50, 763–771.

Nagata, C., Takatsuka, N., & Shimizu, H. (2003). The impact of changes in marital status on the mortalityof elderly Japanese. Annals of Epidemiology, 13, 218–222.

Neale, A. V., Tilley, B. C., & Vernon, S. W. (1986). Marital status, delay in seeking treatment and survivalfrom breast cancer. Social Science & Medicine, 23, 305–312.

Niemi, T. (1979). The mortality of male old-age pensioners following spouse’s death. ScandinavianJournal of Social Medicine, 7, 115–117.

Nilsson, P. M., Nilsson, J.-A., Ostergren, P.-O., & Berglund, G. (2005). Social mobility, marital status, andmortality risk in an adult life course perspective: The Malmö Preventive Project. Scandinavian Journalof Public Health, 33, 412–423.

Nybo, H., Petersen, H. C., Gaist, D., Jeune, B., Andersen, K., McGue, M., . . . Christensen, K. (2003).Predictors of mortality in 2,249 nonagenarians—The Danish 1905-Cohort Survey. Journal of theAmerican Geriatrics Society, 51, 1365–1373.

Nystedt, P. (2002). Widowhood-related mortality in Scania, Sweden during the 19th century. The History ofthe Family, 7, 451–478.

Okamoto, K., Harasawa, Y., Momose, Y., & Sakuma, K. (2007). Risk factors for 6-year mortality by genderin a Japanese elderly population. Archives of Gerontology and Geriatrics, 45, 335–341.

Ortmeyer, C. E. (1974). Variation in mortality, morbidity, and health care by marital status. In C. L. Erhardt& J. E. Berlin (Eds.), Mortality and morbidity in the United States (pp. 159–188). Cambridge, MA:Harvard University Press.

Parkes, C. M., Benjamin, B., & Fitzgerald, R. G. (1969). Broken heart: A statistical study of increasedmortality among widowers. British Medical Journal, 1, 740–743.

Peritz, E., Bialik, O., & Groen, J. J. (1967). Mortality by marital status in the Jewish population of Israel.Journal of Chronic Diseases, 20, 823–828.

Peters, J. L., Sutton, A. J., Jones, D. R., Abrams, K. R., & Rushton, L. (2006). Comparison of two methods todetect publication bias in meta-analysis. Journal of the American Medical Association, 295, 676–680.

Popenoe, D. (1993). American family decline, 1960–1990: A review and appraisal. Journal of Marriageand the Family, 55, 527–555.

Rahman, M. O. (1993). Excess mortality for the unmarried in rural Bangladesh. International Journal ofEpidemiology, 22, 445–456.

Rahman, M. O. (1997). The effects of spouses on the mortality of older people in rural Bangladesh. HealthTransition Review, 7, 1–12.

Rahman, M. O., Foster, A., & Menken, J. (1992). Older widow mortality in rural Bangladesh. SocialScience & Medicine, 34, 89–96.

Rasulo, D., & Christensen, K. (2004). The co-twin control method for assessing the effect of widowhood onmortality. Rostock, Germany: Max Planck Institute for Demographic Research.

Rees, W. D., & Lutkins, S. G. (1967). Mortality of bereavement. British Medical Journal, 4, 13–16.Regidor, E., Calle, M. E., Dominguez, V., & Navarro, P. (2001). Mortality by social and economic

characteristics: The mortality study of the autonomous community of Madrid, Spain (in Spanish).Medicina Clinica, 116, 726–731.

Rosengren, A., Wedel, H., & Wilhelmsen, L. (1989). Marital status and mortality in middle-aged Swedishmen. American Journal of Epidemiology, 129, 54–64.

Ryan, D. H. (1992). Hospital survival time in dementia: A controlled study of the effect of marital status onthe interval between first admission and death. International Journal of Geriatric Psychiatry, 7, 389–397.

Samuelsson, G., & Dehlin, O. (1993). Family network and mortality: Survival chances through the life-span of an entire age cohort. International Journal of Aging & Human Development, 37, 277–295.

Widowhood and Mortality: A Meta-Analysis and Meta-Regression 605

Author's personal copy

Schaefer, C., Quesenberry, C. P., & Wi, S. (1995). Mortality following conjugal bereavement and theeffects of a shared environment. American Journal of Epidemiology, 141, 1142–1152.

Sharar, D. R., Schultz, R., Sharar, A., & Wing, R. R. (2001). The effect of widowhood on weight change,dietary intake, and eating behavior in the elderly population. Journal of Aging and Health, 13, 186–199.

Sheps, M. C. (1961). Marriage and mortality. American Journal of Public Health, 51, 547–555.Shkolnikov, V. M., Jasilionis, D., Andreev, E.M., Jdanov, D. A., Stankuniene, V., & Ambrozaltiene, D. (2007).

Linked versus unlinked estimates of mortality and length of life by education and marital status: Evidencefrom the first record linkage study in Lithuania. Social Science & Medicine, 64, 1392–1406.

Shurtleff, D. (1955). Mortality and marital status. Public Health Reports, 70, 248–252.Shurtleff, D. (1956). Mortality among the married. Journal of the American Geriatrics Society, 4, 654–666.Silverstein, M., & Bengtson, V. L. (1991). Do close parent–child relations reduce the mortality risk of older

parents? Journal of Health and Social Behavior, 32, 382–395.Singh, G. K., & Siahpush, M. (2001). All-cause and cause-specific mortality of immigrants and native born

in the United States. American Journal of Public Health, 91, 392–399.Singh, G. K., & Siahpush, M. (2002). Ethnic-immigrant differentials in health behaviors, morbidity, and

cause-specific mortality in the United States: An analysis of two national data bases. Human Biology,74, 83–109.

Smith, K. R., & Waitzman, N. J. (1994). Double jeopardy: Interaction effects of marital and poverty statuson the risk of mortality. Demography, 31, 487–507.

Smith, K. R., & Zick, C. D. (1996). Risk of mortality following widowhood: Age and sex differences bymode of death. Social Biology, 43, 59–71.

Sorlie, P. D., Backlund, E., & Keller, J. B. (1995). U.S. mortality by economic, demographic, and socialcharacteristics: The National LongitudinalMortality Study. American Journal of Public Health, 85, 949–956.

Spence, N. J. (2006). Reproductive patterns and women’s later life health. Unpublished doctoral disserta-tion, Department of Sociology, Florida State University, Tallahassee, FL.

Stimpson, J. P., Kuo, Y.-F., Ray, L. A., Raji, M. A., & Peek, M. K. (2007). Risk of mortality related towidowhood in older Mexican Americans. Annals of Epidemiology, 17, 313–319.

Stroebe, M. S. (2001). Handbook of bereavement research: Consequences, coping and care. Washington,DC: American Psychological Association.

Stroebe, M., Schut, H., & Stroebe, W. (2007). Health outcomes of bereavement. Lancet, 370, 1960–1973.Stroebe, M. S., Stroebe, W., & Schut, H. (2001). Gender differences in adjustment to bereavement: An

empirical and theoretical review. Review of General Psychology, 5, 62–83.Subramanian, S. V., Elwert, F., & Christakis, N. A. (2008). Widowhood and mortality among the elderly:

The modifying role of neighborhood concentration of widowed individuals. Social Science & Medi-cine, 66, 873–884.

Susser, M. (1981). Widowhood: A situational life stress or a stressful life event? American Journal ofPublic Health, 71, 793–795.

Thierry, X. (1999). Risks of mortality and excess mortality during the first ten years of widowhood.Population, 54, 177–204.

Tomassini, C., Billari, F. C., Rosina, A., & Christensen, K. (2001). Born together—Die together. Live together—Die together. The role of the partner and the co-twin on longevity at very old ages. Genus, 62, 63–82.

Vallin, J., & Nizard, A. (1977). Mortality by marital status: Marriage selection or marriage protection.Population (French ed.), 32, 95–125.

van Poppel, F. W. A., & Joung, I. M. A. (2001). Long-term trends in marital status mortality differences inthe Netherlands 1850–1970. Journal of Biosocial Science, 33, 279–303.

Villingshoj, M., Ross, L., Thomsen, B. L., & Johansen, C. (2006). Does marital status and altered contact withthe social network predict colorectal cancer survival? European Journal of Cancer, 42, 3022–3027.

Voges, W. (1996). Unequal requirements for longevity—Social determinants of mortality in longitudinalperspective (in German). Zeitschrift Fur Gerontologie Und Geriatrie, 29, 18–22.

Young, M., Benjamin, B., & Wallis, C. (1963). The mortality of widowers. Lancet, 2, 454–456.Zajacova, A. (2006). Education, gender, and mortality: Does schooling have the same effect on mortality

for men and women in the US? Social Science & Medicine, 63, 2176–2190.Zhang, J., & Yu, K. F. (1998). What’s the relative risk? A method of correcting the odds ratio in cohort

studies of common outcomes. Journal of the American Medical Association, 280, 1690–1691.Zick, C. D., & Smith, K. R. (1991a). Marital transitions, poverty, and gender differences in mortality.

Journal of Marriage and the Family, 53, 327–336.Zick, C. D., & Smith, K. R. (1991b). Patterns of economic change surrounding the death of a spouse.

Journal of Gerontology, 46, S310–S320.

606 E. Shor et al.

Author's personal copy